Problem 315: Digital root clocks

(see projecteuler.net/problem=315)

clocks

Sam and Max are asked to transform two digital clocks into two "digital root" clocks.
A digital root clock is a digital clock that calculates digital roots step by step.

When a clock is fed a number, it will show it and then it will start the calculation, showing all the intermediate values until it gets to the result.
For example, if the clock is fed the number 137, it will show: "137" → "11" → "2" and then it will go black, waiting for the next number.

Every digital number consists of some light segments: three horizontal (top, middle, bottom) and four vertical (top-left, top-right, bottom-left, bottom-right).
Number "1" is made of vertical top-right and bottom-right, number "4" is made by middle horizontal and vertical top-left, top-right and bottom-right. Number "8" lights them all.

The clocks consume energy only when segments are turned on/off.
To turn on a "2" will cost 5 transitions, while a "7" will cost only 4 transitions.

Sam and Max built two different clocks.

Sam's clock is fed e.g. number 137: the clock shows "137", then the panel is turned off, then the next number ("11") is turned on, then the panel is turned off again and finally the last number ("2") is turned on and, after some time, off.
For the example, with number 137, Sam's clock requires:
"137": (2 + 5 + 4) * 2 = 22 transitions ("137" on/off).
"11": (2 + 2) * 2 = 8 transitions ("11" on/off).
"2": (5) * 2 = 10 transitions ("2" on/off).
For a grand total of 40 transitions.

Max's clock works differently. Instead of turning off the whole panel, it is smart enough to turn off only those segments that won't be needed for the next number.
For number 137, Max's clock requires:
"137": 2 + 5 + 4 = 11 transitions ("137" on)
7 transitions (to turn off the segments that are not needed for number "11").
"11": 0 transitions (number "11" is already turned on correctly)
3 transitions (to turn off the first "1" and the bottom part of the second "1"; the top part is common with number "2").
"2": 4 transitions (to turn on the remaining segments in order to get a "2")
5 transitions (to turn off number "2").
For a grand total of 30 transitions.

Of course, Max's clock consumes less power than Sam's one. The two clocks are fed all the prime numbers between A = 10^7 and B = 2 * 10^7.

Find the difference between the total number of transitions needed by Sam's clock and that needed by Max's one.

My Algorithm

I encode all lit segments as bitmasks in Segments. Counting the set bits (popcnt) is equivalent to counting the lit segments.
There are at most 8 digits, hence a 64 bit integer represents all segments of a number.

Sam's algorithm can be found in sam: switching on and off all segments "costs" 2x popcnt(getSegments(x)).

Max's algorithm (see max) is a bit more complicated: I have to compare all lit segments of the previous number to the lit segments of the current number.
When I XOR both bitmasks then I get a new bitmasks with all segments that change (either have to be turned off or turnd on).
Counting the set bits in the XORed value tells me the number of transitions.
Before the first and after the last number all lit segments transition from and to the "empty" state.

Both routines work recursively because I hoped to improve the execution time by caching / memoizing results:
After the first step the digitSum is below 70. However, the simple memoization in sam didn't help much
and therefore I didn't even attempt to code something similiar in max.

Note

popcnt was named after the super-fast CPU instruction which counts bits.
My G++ compiler shows the result after 0.12 seconds when that intrinsic can be used but needs 0.17 seconds with my own alternative code
(which still relies on tricky bit manipulation, see bits.stephan-brumme.com).

My code

… was written in C++11 and can be compiled with G++, Clang++, Visual C++. You can download it, too.

#include <iostream>
#include <vector>
 
// represent digits as bitmasks, lit segments are 1
const unsigned char Segments[10] =
{
// assume the following numbering scheme for the seven segments:
// 000
// 1 2
// 1 2
// 333
// 4 5
// 4 5
// 666
(1 << 0) | (1 << 1) | (1 << 2) | (0 << 3) | (1 << 4) | (1 << 5) | (1 << 6), // digit 0 = 0x77
(0 << 0) | (0 << 1) | (1 << 2) | (0 << 3) | (0 << 4) | (1 << 5) | (0 << 6), // digit 1 = 0x24
(1 << 0) | (0 << 1) | (1 << 2) | (1 << 3) | (1 << 4) | (0 << 5) | (1 << 6), // digit 2 = 0x5d
(1 << 0) | (0 << 1) | (1 << 2) | (1 << 3) | (0 << 4) | (1 << 5) | (1 << 6), // digit 3 = 0x6d
(0 << 0) | (1 << 1) | (1 << 2) | (1 << 3) | (0 << 4) | (1 << 5) | (0 << 6), // digit 4 = 0x2e
(1 << 0) | (1 << 1) | (0 << 2) | (1 << 3) | (0 << 4) | (1 << 5) | (1 << 6), // digit 5 = 0x6b
(1 << 0) | (1 << 1) | (0 << 2) | (1 << 3) | (1 << 4) | (1 << 5) | (1 << 6), // digit 6 = 0x3b
(1 << 0) | (1 << 1) | (1 << 2) | (0 << 3) | (0 << 4) | (1 << 5) | (0 << 6), // digit 7 = 0x27
(1 << 0) | (1 << 1) | (1 << 2) | (1 << 3) | (1 << 4) | (1 << 5) | (1 << 6), // digit 8 = 0x7f
(1 << 0) | (1 << 1) | (1 << 2) | (1 << 3) | (0 << 4) | (1 << 5) | (1 << 6) // digit 9 = 0x6f
};
 
 
// ---------- standard prime sieve from my toolbox ----------
 
// odd prime numbers are marked as "true" in a bitvector
std::vector<bool> sieve;
 
// return true, if x is a prime number
bool isPrime(unsigned int x)
{
// handle even numbers
if ((x & 1) == 0)
return x == 2;
 
// lookup for odd numbers
return sieve[x >> 1];
}
 
// find all prime numbers from 2 to size
void fillSieve(unsigned int size)
{
// store only odd numbers
const unsigned int half = (size >> 1) + 1;
 
// allocate memory
sieve.resize(half, true);
// 1 is not a prime number
sieve[0] = false;
 
// process all relevant prime factors
for (unsigned int i = 1; 2 * i*i < half; i++)
// do we have a prime factor ?
if (sieve[i])
{
// mark all its multiples as false
unsigned int current = 3 * i + 1;
while (current < half)
{
sieve[current] = false;
current += 2 * i + 1;
}
}
}
 
// ---------- problem solution ----------
 
// return sum of all digits
unsigned int digitSum(unsigned int x)
{
unsigned int result = 0;
while (x > 0)
{
result += x % 10;
x /= 10;
}
return result;
}
 
// convert number to its bitmasks, 8 bits per digit => x must not exceed 99999999
unsigned long long getSegments(unsigned int x)
{
unsigned long long result = 0;
unsigned int shift = 0;
while (x > 0)
{
unsigned long long current = Segments[x % 10];
result |= current << shift;
x /= 10;
shift += 8;
}
return result;
}
 
// count bits which are set to 1
unsigned int popcnt(unsigned long long x)
{
#ifdef __GNUC__
return __builtin_popcount(x);
#endif
 
unsigned int result = 0;
while (x > 0)
{
// clear right-most bit
x &= x - 1;
result++;
}
return result;
}
 
// process all steps with strategy of Sam, return number of transitions
unsigned int sam(unsigned int x)
{
// memoize (runs quite fast without, too)
static std::vector<unsigned int> cache(100, 0);
if (x < cache.size() && cache[x] > 0)
return cache[x];
 
// display was clear
auto segments = getSegments(x);
// switch (intermediate) number on and off
auto result = 2 * popcnt(segments);
 
// need further iterations ?
if (x > 9)
result += sam(digitSum(x));
 
if (x < cache.size())
cache[x] = result;
return result;
}
 
// process all steps with strategy of Max, return number of transitions
unsigned int max(unsigned int x, unsigned long long previousSegments = 0)
{
auto segments = getSegments(x);
auto transitions = segments ^ previousSegments;
auto result = popcnt(transitions);
 
if (x > 9)
// morph to next number
result += max(digitSum(x), segments);
else
// fade out
result += popcnt(segments);
 
return result;
}
 
int main()
{
// search range
unsigned int from = 10000000;
unsigned int to = 20000000;
std::cin >> from >> to;
 
// for live input only
if (from > to)
return 1;
 
// generate prime sieve
fillSieve(to);
 
// count all digits of Sam and Max
unsigned long long sumSam = 0;
unsigned long long sumMax = 0;
for (auto i = (from | 1); i <= to; i += 2) // odd numbers only
{
if (!isPrime(i))
continue;
 
sumSam += sam(i);
sumMax += max(i);
}
 
// Sam >= Max, display the difference
std::cout << sumSam - sumMax << std::endl;
return 0;
}

This solution contains 27 empty lines, 39 comments and 4 preprocessor commands.

Interactive test

You can submit your own input to my program and it will be instantly processed at my server:

Input data (separated by spaces or newlines):
Note: Enter the begin and the end of the range to be analyzed.

This is equivalent to
echo "137 137" | ./315

Output:

(please click 'Go !')

Note: the original problem's input 10000000 20000000 cannot be entered
because just copying results is a soft skill reserved for idiots.

(this interactive test is still under development, computations will be aborted after one second)

Benchmark

The correct solution to the original Project Euler problem was found in 0.13 seconds on a Intel® Core™ i7-2600K CPU @ 3.40GHz.
Peak memory usage was about 3 MByte.

(compiled for x86_64 / Linux, GCC flags: -O3 -march=native -fno-exceptions -fno-rtti -std=c++11 -DORIGINAL)

See here for a comparison of all solutions.

Note: interactive tests run on a weaker (=slower) computer. Some interactive tests are compiled without -DORIGINAL.

Changelog

July 29, 2017 submitted solution
July 29, 2017 added comments

Difficulty

20% Project Euler ranks this problem at 20% (out of 100%).

Heatmap

green problems solve the original Project Euler problem and have a perfect score of 100% at Hackerrank, too.
yellow problems score less than 100% at Hackerrank (but still solve the original problem).
gray problems are already solved but I haven't published my solution yet.
blue problems are solved and there wasn't a Hackerrank version of it at the time I solved it or I didn't care about it because it differed too much.
red problems are solved but exceed the time limit of one minute or the memory limit of 256 MByte.

Please click on a problem's number to open my solution to that problem:

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50
51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75
76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100
101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125
126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150
151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175
176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200
201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225
226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250
251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275
276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300
301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325
326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350
351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375
376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400
401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425
426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450
451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475
476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500
501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525
526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550
The 235 solved problems (level 9) had an average difficulty of 29.1% at Project Euler and
I scored 13,183 points (out of 15300 possible points, top rank was 17 out of ≈60000 in August 2017) at Hackerrank's Project Euler+.
Look at my progress and performance pages to get more details.

My username at Project Euler is stephanbrumme while it's stbrumme at Hackerrank.

more about me can be found on my homepage, especially in my coding blog.
some names mentioned on this site may be trademarks of their respective owners.
thanks to the KaTeX team for their great typesetting library !